This is a follow-up to the discussion in this note.
What I want to do
I want to use glow in the dark paint to calibrate camera color sensors. In this experiment I wanted to test whether it would be feasible to do (i.e. is the paint bright enough for the camera to pick up).
My attempt and results
I photographed the green paint found in Palmer 12-Pot Fluorescent Glow in the Dark Set applied to a white sheet of paper. Other colors did not appear to be as bright so I thought green would be a good starting point.
I charged the paint using sunlight. I did not make any effort to ensure the maximum charge. It is possible that I could have gotten a better charge by using a more intense light source, angling the paper to face the sun directly and charging for longer periods of time. This is a photo of the paint while it was charging:
Next I photographed the paper using a Samsung Galaxy S3 with +2 exposure in a very dark room (there was some ambient light from outside the door):
I am unable to make out any green glow in the original picture, however, I when applied Aviary's Auto-enhance filter to it green blobs appeared:
Questions:
- Is the paint bright enough to effectively do calibration? I think it might not be since the intensity of each channel might only range from 0-2.
- Are there any camera apps that allow longer exposure times? In the live preview mode?
- Could I spatially average the pixel intensities to get a more precision in my measurements?
Next steps:
- Try to get the paint to glow with greater intensity.
- See if other colors will show up.
- Create a rainbow glow-in-the-dark calibration card and see if I can use it to calibrate two different cameras so the colors in a reference scene match up.
3 Comments
I think there are some apps which allow long exposures (Camera FV-5). I guess the calibration photo of the paint would have to use the same settings (e.g., ISO, white balance) as the photo you want to calibrate, because those settings will influence color representation. That might be possible in some situations but not in others. But maybe to calibrate two cameras with each other that is not critical, especially if the color representation of the two cameras vary due to settings in the same way. But if they vary in the same way, maybe they don't need to be calibrated?
Is this a question? Click here to post it to the Questions page.
Reply to this comment...
Log in to comment
Real long-exposure would be great; I think Camera FV-5 uses software based long exposure -- it just adds together a lot of half-second exposures, so if you are getting black (or noise) in each, they'll just sum to a lot of noise.
http://www.camerafv5.com/faq.php#long-exposure-resolution
If it actually worked, it'd be extra useful for smartphone spectroscopy.
Reply to this comment...
Log in to comment
I think the Camera FV-5 Approach of adding multiple image might work. The noise, if it's randomly distributed, should just sum to gray, while the illuminated pixels should accumulate more. Although, the illuminated pixels might not accumulate in proportion to their intensity. If a pixel with 0.8 illumination varies between 0 and 1 so that it is 1 80% of the time and 0 20% of the time the accumulation might be proportional. Another issues is that I suspect Camera FV-5 sums the images in a way that causes the illuminated pixels to be lost due to rounding error, because I haven't been able to get them to show up at all with its long exposure method.
I've discovered that the glow in the dark paint is actually quite visible in my camera's night mode and low light mode for about 30 seconds after being charged by being held approximately 5 inches from a light bulb. Since the paint discharges so quickly brightness calibration might be tricky. I think it could still be automated. The paints seem to discharge at different rates, and that could be exploited to determine the absolute brightness because the brightness of the paints relative to each-other might indicate how long they have been discharging.
Reply to this comment...
Log in to comment
Login to comment.